Department of Diagnostic Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, 17 Panjiayuan Nanli, Chaoyang District, Beijing 100021, China (R.Z., L.W., S.C., W.P., X.L., L.L., H.Z.).
Department of Pharmaceutical Diagnosis, GE Healthcare, Life Sciences, Beijing, China (S.W.).
Acad Radiol. 2024 Apr;31(4):1367-1377. doi: 10.1016/j.acra.2023.09.007. Epub 2023 Oct 4.
To develop and validate a nomogram based on intratumoral and peritumoral radiomics signatures for pretreatment prediction of distant metastasis-free survival (DMFS) in patients after neoadjuvant chemoradiotherapy (NCRT) with locally advanced rectal cancer (LARC).
This retrospective study included 230 patients (161 training cohort; 69 validation cohort) with LARC who underwent NCRT and surgery. Radiomics features were extracted on T2-weighted images from gross tumor volume (GTV) and volumes of 4-mm, 6-mm, and 8-mm peritumoral regions (PTV, PTV, and PTV). The least absolute shrinkage and selection operator (LASSO)-Cox analysis were used for features selection and models construction. The performance of each model in predicting DMFS was evaluated by the Concordance index (C-index) and time-independent receiver operating characteristic curve (ROC).
The PTV radiomics model demonstrated superior performance compared to the PTV and PTV radiomics models, with C-indexes of 0.750 and 0.703 in the training and validation cohorts, respectively. The nomogram was constructed by integrating the GTV radiomics signature, PTV radiomics signature, and relevant clinical characteristics, including CA19-9 level, clinical T stage, and clinical N stage. The nomogram achieved C-indexes of 0.831 and 0.748, with corresponding AUCs of 0.872 and 0.808 for 5-year DMFS in the training and validation cohorts, respectively. Kaplan-Meier analysis revealed that a cut-off value of 1.653 effectively stratified patients into high- and low-risk groups for DM (P < 0.001).
The intra-peritumoral radiomics nomogram is a favorable tool for clinicians to develop personalized systemic treatment and intensive follow-up strategies to improve patient prognosis.
为了预测接受新辅助放化疗（NCRT）的局部晚期直肠癌（LARC）患者无远处转移生存（DMFS），我们开发并验证了一种基于肿瘤内和肿瘤周围放射组学特征的列线图。
本回顾性研究纳入了 230 例接受 NCRT 联合手术治疗的 LARC 患者（161 例训练队列；69 例验证队列）。在 GTV 和 4mm、6mm、8mm 肿瘤周围区域（PTV、PTV 和 PTV）的 T2 加权图像上提取放射组学特征。采用最小绝对收缩和选择算子（LASSO）-Cox 分析进行特征选择和模型构建。通过一致性指数（C-index）和时间独立的接收者操作特征曲线（ROC）评估每个模型预测 DMFS 的性能。
与 PTV 和 PTV 放射组学模型相比，PTV 放射组学模型具有更好的性能，其在训练和验证队列中的 C-index 分别为 0.750 和 0.703。通过整合 GTV 放射组学特征、PTV 放射组学特征以及 CA19-9 水平、临床 T 分期和临床 N 分期等相关临床特征，构建了列线图。该列线图在训练和验证队列中预测 5 年 DMFS 的 C-index 分别为 0.831 和 0.748，对应的 AUC 分别为 0.872 和 0.808。Kaplan-Meier 分析显示，分界值为 1.653 可有效将患者分为 DMFS 高风险和低风险组（P<0.001）。
肿瘤内-肿瘤周围放射组学列线图是临床医生制定个体化全身治疗和强化随访策略以改善患者预后的有利工具。
